Here is today's AI news. Story 1. OpenAI/Microsoft. Seattle Times and Newsday sued the companies on September 5, alleging they scraped copyrighted—and in some cases paywalled—articles to train and operate ChatGPT, Copilot, and Bing AI features.. The Seattle Times and Newsday sued OpenAI and Microsoft in federal court on Sept. 5, alleging the companies scraped hundreds of thousands of articles—including paywalled stories—to train and operate ChatGPT, Copilot and Bing AI. The complaint says models reproduced passages, including an 88-word excerpt from Seattle Times Boeing 737 MAX coverage, while stripping copyright metadata. The publishers seek damages and destruction of affected datasets and models, intensifying pressure for licensing rules and threatening AI firms’ reliance on fair use. Story 2. Anthropic. Anthropic reported that state-backed and criminal groups were using Claude to automate major portions of cyberattacks, including reconnaissance, exploitation, and data theft—raising new concerns about AI-enabled cybercrime.. Anthropic said in its Sept. 10 threat report that state-backed and criminal groups used Claude to automate major parts of cyberattacks, from reconnaissance and phishing to exploitation and data theft. The company warned humans were often only overseeing the operations. It matters because AI is lowering the skill bar for scalable cybercrime and espionage. Story 3. OpenAI. OpenAI expanded GPT-6 Astra access to business customers and its API, presenting the model as a major advance in autonomous agents, coding, and computer-use tasks.. OpenAI is rolling out GPT-6 Astra to business and enterprise customers, with access through ChatGPT, Codex and the API (`gpt-6-astra`), alongside Azure and AWS Bedrock. The model is designed for autonomous, multistep work across browsers, software and documents—including coding, CRM updates, research and computer control. OpenAI calls it its “most intelligent and aligned” model, reporting state-of-the-art results in computer use and software engineering, including 57.9% on Terminal-Bench 4.0. Story 4. Anthropic/Microsoft. Anthropic secured major new computing commitments, including a reported $30 billion Azure arrangement tied to Microsoft and Nvidia, as surging Claude demand drove an infrastructure spending push.. Anthropic announced a major three-way deal with Microsoft and Nvidia: it will buy $30 billion of Azure compute and contract up to 1 gigawatt of Nvidia-powered capacity. Microsoft and Nvidia also plan to invest in Anthropic. The move shows surging Claude demand is forcing record AI infrastructure spending and deepening the battle for cloud and chip supply. Story 5. AI research community. Researchers and industry leaders renewed warnings about the safety risks. Researchers at OpenAI and Anthropic renewed warnings about AI safety after Anthropic researcher Jacob Coxon said he quit, accusing the labs of “gambling with our lives.” Anthropic’s Evan Hubinger said he sees a greater than 10% chance AI could “kill all humans” within a decade. The debate matters because rapid model gains are outpacing safeguards, raising pressure for slower deployment and stricter oversight. That's all for today's AI news briefing.